Empirical analysis of climate-based malaria suitability models: do knowledge-based thresholds (derived without disease case data) actually correlate with observed malaria incidence?
The analysis applies Uganda DHIS2 malaria suitability thresholds (temperature 20–30°C, rainfall ≥100 mm/month, humidity 50–80%) to Lao PDR admin-1 monthly data (1998–2010) across 17 provinces.
suitability/ — SuitabilityModel and ThresholdComponent
configs/ — Model config (chap_malaria.json)
analysis/ — Analysis modules (correlation, lag, sensitivity, etc.)
report.py — CHAP report entry point
visualize.py — Plot functions and PDF generation
The model is CHAP-compatible and can be run via chap report on any dataset with columns time_period, location, disease_cases, rainfall, mean_temperature, humidity. CHAP fetches climate covariates from DHIS2 automatically based on required_covariates in MLproject.
Note: relative humidity is not always included in CHAP's default covariate set.